Robert Baldwin From regulation to behaviour change: giving ...eprints.lse.ac.uk › 64343 › 1 ›...
Transcript of Robert Baldwin From regulation to behaviour change: giving ...eprints.lse.ac.uk › 64343 › 1 ›...
Robert Baldwin
From regulation to behaviour change: giving nudge the third degree Article (Accepted version) (Refereed)
Original citation: Baldwin, Robert (2014) From regulation to behaviour change: giving nudge the third degree. The Modern Law Review, 77 (6). pp. 831-857. ISSN 0026-7961
DOI: 10.1111/1468-2230.12094 © 2014 The Author This version available at: http://eprints.lse.ac.uk/64343/ Available in LSE Research Online: November 2015 LSE has developed LSE Research Online so that users may access research output of the School. Copyright © and Moral Rights for the papers on this site are retained by the individual authors and/or other copyright owners. Users may download and/or print one copy of any article(s) in LSE Research Online to facilitate their private study or for non-commercial research. You may not engage in further distribution of the material or use it for any profit-making activities or any commercial gain. You may freely distribute the URL (http://eprints.lse.ac.uk) of the LSE Research Online website. This document is the author’s final accepted version of the journal article. There may be differences between this version and the published version. You are advised to consult the publisher’s version if you wish to cite from it.
1
From Regulation to Behaviour Change: Giving Nudge the Third Degree
Robert Baldwin*
Key Words: Regulation; Nudge; Behaviour Change. Behaviour change strategies such as ‘nudge’ have become hugely popular with administrations on both
sides of the Atlantic. Nudging, however, is a practice that raises both conceptual and controversial
issues and these have to be addressed in examining the conditions under which nudging can be used
effectively and acceptably. Hitherto there has been a lack of conceptual clarity in many debates about
nudge but a key to understanding nudge-related issues is to distinguish between three very different
degrees of nudge. These three degrees raise different, and identifiable, concerns and it is possible to
assess the extent to which these worries can be responded to in positive terms. The compatibility of
nudging with other control devices cannot be assumed and, when contemplating nudging, it is essential
to be clear and open about the philosophical basis for such action as well as to be aware that different
modes of intervention may operate with clashes of logic that threaten not only effectiveness but also
the serving of representative and ethical ends.
___________________________________________________________________________________
‘Nudging’ involves structuring the choices that people make so as to lead them
towards particular outcomes. Placing fruit next to the supermarket till, for example,
gives a nudge towards healthy eating. Such ‘behaviour change’ strategies have
become hugely popular with administrations on both sides of the Atlantic. In the UK,
David Cameron set up a Behavioural Insights Team (or ‘Nudge Unit’) at the centre of
UK government in 20101 so as to foster alternatives to traditional regulation and to
move towards less-restrictive and lower-cost controls of behaviour. The governmental
endorsement of nudging has not, however, been based on clear positions regarding the
nature of ‘nudge’ or the role of nudge in the array of state control devices.
This article seeks to address these deficiencies and presents three central
arguments. First, that, before nudging is further embraced, it is necessary to think
much more clearly about the concept of nudge and to distinguish between the
different degrees of nudge that can be applied to targets. Second, that different
degrees of nudge can present distinct issues of effectiveness and can give rise to
separate concerns of a representative and ethical nature. Third, that a number of
important issues are easily overlooked if nudge is seen as just another mode of
influence in the toolbox of state controls.
* Professor of Law, London School of Economics and Political Science. The author is
grateful for the comments of Vanessa Finch, Veerle Heyvaert, Andrew Lang, Martin Lodge,
Henry Rothstein and the anonymous referees.
1 The Behavioural Insights Team (BIT, or 'Nudge Unit') was set up in July 2010 with a remit
to find innovative ways of encouraging, enabling and supporting people to make better
choices for themselves. Professor Richard Thaler was appointed as an external expert to the
BIT - See more at: http://blogs.cabinetoffice.gov.uk/behavioural-insights-
team/author/behavioural-insights-team/#sthash.mxq3kUis.dpuf
In early 2014 the nudge unit was part-privatised by sale to a charity and to employees. See:
http://www.bbc.co.uk/news/uk-politics-26030205
2
The article starts by outlining the rise of ‘nudge’ before reviewing the concept of
nudge and distinguishing between three different degrees of nudge. It then explores
the limits to the effectiveness of the nudge strategy and the representational and
ethical problems posed by the different degrees of nudge. The conditions for effective
and acceptable use of nudge are then investigated before the role of nudge in relation
to other intervention methods is explored. Emphasis will be placed on the dangers
involved in seeing nudge as simply as an extra option in the toolbox of control
devices. Finally, conclusions will be offered on the feasibility of constructing a
conceptual and ethical basis for making choices regarding this fashionable
intervention tool.
THE RISE OF NUDGE
The idea of ‘nudge’ is commonly associated with the book of that title published by
Richard Thaler and Cass Sunstein in 2008.2 The two authors define a ‘nudge’ as: ‘…
any aspect of the choice architecture that alters people’s behaviour in a predictable
way without forbidding any options or significantly changing their economic
incentives. To count as a nudge, the intervention must be easy and cheap to avoid.’3 In
this conception the ‘choice architecture’ is the environment that frames an
individual’s choice. Thus: ‘Putting the fruit at eye level counts as a nudge. Banning
junk food does not.’4
The proponents of nudge build on the well-established insights of cognitive
psychology and behavioural economics to contend that control systems need to take
on board the bounded rationality of citizens when they make daily decisions.5
Individuals are seen as constrained by limitations of information, cognitive capacity
and self-control and as tending to rely on a series of heuristics and shorthand methods
of using information.6 As a result, they often make poor decisions that do not serve
their welfare. Nudging holds that citizens’ choices tend to be influenced by a host of
framing factors – such as the ways in which options are presented – and it advocates
structuring choice architectures so that it is easier for consumers or others (such as
regulated firms) to act in ways that are beneficial to them.7
At the heart of nudging is a philosophy entitled ‘libertarian paternalism’.8 Thus,
nudge is said to possess a paternalistic dimension in stimulating choices that are seen
2 R. Thaler and C. Sunstein, Nudge (New Haven: Yale UP, 2008; revised as Penguin 2009).
See also the two authors’ ‘Libertarian Paternalism’ (2003) 93 American Economic Review
175-179; ‘Libertarian Paternalism is Not an Oxymoron (2003) 70 University of Chicago Law
Review 1159-1202’. 3 Thaler and Sunstein 2008, n 2 above, 6 (page references in this article are to the revised
Penguin edition of 2009). 4 ibid. p.8.
5 On such insights see e.g. A. Tversky and D. Kahneman, ‘Judgement Under Uncertainty:
Heuristics and Biases’ (1974) 185 Science 1124-1131; H. Simon, Administrative Behaviour
(Fourth Edition) (New York: The Free Press, 1997). 6 Thaler and Sunstein 2008, n 2 above, Chapter 1) note many factors that lead to mistakes in
human judgement, such as: optimism and over-confidence; loss aversion; a status quo bias;
framing; inertia; inattention and error; anchoring; availability; and representativeness. 7 In their first example of a nudge, Thaler and Sunstein posit that the food in a school cafeteria
is arranged so that students are likely to make the most healthy choices Thaler and Sunstein
2008, n 2 above, 5. 8 Thaler and Sunstein 2008, n 2 above, 5-6.
3
as welfare enhancing for the individual but it combines this with a libertarian aspect in
so far as it purports to leave the target person or firm free to choose to take the non-
sensible course of action.
Leaving aside incentive regimes, Thaler and Sunstein give examples of at least
seven nudge tools.9 Defaults encourage the healthy choice by making this the given
selection unless active steps are taken to avoid this outcome – steps that will not be
encouraged by the natural tendency to human inertia. A prime example of a default is
a presumption that all citizens consent to be organ donors unless they register their
unwillingness to donate (which, as noted, Thaler and Sunstein state, they should be
able to do easily). Similarly, employees can be enrolled automatically onto a pension
scheme unless they actively opt out. Persuasive, campaigning and counselling
strategies can influence and shape decision-making. Design approaches can be used
– as where the main office doors lead directly to the stairs, not the lifts; or the
departmental smoking zone is placed at a distance from the work area. Commitments
to healthy activities, such as eating less or charitable giving can be stimulated so as to
encourage delivery on good intentions. Transactional shortcuts can be provided so
that, for example, a special credit card makes charitable donations easy to make and
record. Exemptions, similarly, can be used to spare people from procedures if they do
the sensible thing – as where motorcycle helmet users do not have to obtain a licence.
Information mechanisms can be deployed to inform people of pitfalls or errors, or
to present options in ways that foster healthy decisions, or to facilitate corrective
actions. Information can also be given on social norms or other people’s performance
(e.g. the percentage of citizens who have already completed a tax return)10
.
Information can also be mapped and structured (especially in relation to complex
issues) so as to provide citizens with easier routes to the sensible decision. Warnings
and reminders can be given so as to discourage unwise actions such as the sending
of intemperate or uncivil emails.11
As noted, the Cameron administration set up the Behavioural Insights Team (the
‘Nudge Unit’) as part of a movement away from command-based regulation and fiscal
controls in favour of ‘smarter’ strategies that would use behavioural insights rather
than rules or financial incentives to influence behaviour. This drift was seen as
continuing the deregulatory thrust of the Better Regulation Executive and as an
espousing of ‘non-regulatory means of achieving behaviour change’.12
As for the
merits of a ‘non-regulatory approach’, the Minister for Government Policy, Oliver
Letwin MP, stressed four factors: effectiveness, cost-effectiveness, less rigid
impositions on individuals and reduced burdens on business.
Since its inception the BIT has investigated the potential of nudges to improve
practices relating to such matters as: organ donor registrations, charitable donations,
energy saving and smoking.13
9 Incentive strategies with economic effects are not nudges as defined by Thaler and Sunstein
2008, n 2 above, 8. For a dozen examples of nudges see Thaler and Sunstein (2008) Chapter
14. 10
See e.g. R. Cialdini, ‘Activating and Aligning Social Norms for Persuasive Impact’ (2006)
Social Influence 3-15. 11
The example provided at Thaler and Sunstein 2008, n 2 above, 233-234. 12
House of Lords Science and Technology Select Committee Second Report of Session 2010-
12 Behaviour Change (HL Paper 179, 2011) (hereafter ‘House of Lords Select Committee,
2011’) para. 5.3 quoting the evidence of the head of the BIT, Dr David Halpern. 13
For details of the BIT’s work see its website:
https://www.gov.uk/government/organisations/behavioural-insights-team
4
THE CONCEPT OF NUDGE
Understanding what nudging involves is a pre-condition for assessing the
effectiveness and acceptability of nudge as an intervention strategy – or for evaluating
any potential use of nudge in a specific context. If the concept of ‘nudge’ is unclear or
encompasses a number of very different devices this complicates any attempt to
assess its potential performance.
A problem here is that Thaler and Sunstein adopt a conception of the nudge that
is both broad and contested by some commentators. Thus, the two authors include
within nudging a number of examples of simple information provision (e.g.
requirements that cigarette packs display information on the risks of smoking; or
demands that firms inform employees of work hazards; or suggestions that red
warning lights be used to inform of needs to change air filters14
). They also cloud
matters by citing, as examples of nudges, some uses of economic incentives (e.g. to
limit pregnancies15
) which contradicts their definitional exclusion of fiscal measures
from the category of nudges.16
On the matter of simple information supply, some commentators would depart
from Thaler and Sunstein and omit these activities from the category of ‘nudges’.17
Thus Peter John and colleagues have (like others) distinguished between ‘nudges’ and
actions that are taken so as to foster deliberative decision-making processes.18
‘Nudges’, thus conceived, operate by working within the domain of ‘automatic’ (or
‘System 1’) responses since the essence of nudging is to control individuals by
exploiting their cognitive and emotional limitations rather than by seeking to improve
their capacities to exercise informed, rational and conscious choices. In contrast,
interventions that are based on a ‘think’ strategy are designed to encourage
‘reflective’ (or ‘System 2’) decision procedures.19
‘Think’ strategies will enhance
rationality by supplying information that will be used in reflective decision-making
(‘These cigarettes may harm your health’). ‘Simple’ information supply, from this
perspective, is thus inconsistent with nudging as it stimulates conscious decision-
making.
Those who disagree with Thaler and Sunstein on this point reinforce their
position by arguing that simple information provision is not, in any event, covered by
the idea of ‘libertarian paternalism’ because actions that are taken so as to enhance an
individual’s exercise of ‘reflective’, rational choice, and which respect the decision-
14
See Thaler and Sunstein 2008, n 2 above, 200-201; 232-233 and D. Hausman and B.
Welch, ‘Debate: To Nudge or Not to Nudge’ (2010) 18 The Journal of Political Philosophy
123-136, 127. 15
See Thaler and Sunstein 2008, n 2 above, 232. 16
See: British Medical Journal, Editorial ‘One Nudge Forward, Two Steps Back’ (2011) 342
BMJ d401. 17
P. John, G. Smith and G. Stoker, ‘Nudge Nudge, Think Think: Two Strategies for
Changing Civic Behaviour’ (2009) 80 The Political Quarterly 361-370. The House of Lords
Select Committee, 2011, similarly took the view, contra Thaler and Sunstein, that: ‘Nudges
prompt choices without getting people to consider their options consciously, and therefore do
not include openly persuasive interventions such as media campaigns and the straightforward
provision of information.’ (para. 2.9). See also L. Bovens, ‘The Ethics of Nudge’ in T.
Grune-Yanoff and S. Hansson (eds.) Preference Change (Berlin: Springer, 2008). 18
John, Smith and Stoker ibid. 19
See Thaler and Sunstein 2008, n 2 above, 21. More generally see D. Kahneman, Thinking,
Fast and Slow (London: Allen Lane, 2011).
5
making autonomy of the individual, are not paternalistic at all – there is no
substitution of the judgement of the nudger for that of the individual who is nudged.20
Thaler and Sunstein have, nevertheless, been followed by many policy-makers in
including ‘simple information’ and other ‘think’ devices alongside ‘System 1’
influences within the category of nudges.21
THREE DEGREES OF NUDGE
The prevalence of the latter, highly inclusive, notion of nudge means that
distinctions have to be drawn between different degrees of nudge if there is to be
conceptual clarity on nudging. Nudges, it is argued here, vary in the degree to which
they impact on the autonomy of individuals as decision-makers and it is useful to
distinguish between three levels of such impact. This is because distinct sets of
representative, ethical and practical issues are raised by these three degrees of nudge.
It should be noted that the degree of a nudge is a separate matter from the
particular nudging tool that is deployed (be that a default rule, an exercise in
information supply or some other approach). Thus, a tool such as supplying
information may be used, as will be demonstrated, for first, second, or third degree
nudging.
In the framework proposed here, ‘First Degree nudges’ respect the decision-
making autonomy of the individual and enhance reflective decision-making. Typical
First Degree nudges involve the supply of simple information to individuals or the
imparting of reminders (‘There are three weeks left to complete the tax return.’). Such
nudges can be distinguished from two other, more serious, kinds of intervention.
A ‘Second Degree nudge’ typically builds on behavioural or volitional limitations
so as to bias a decision in the desired direction. Thus, a default rule with an opt-out
can be used to shape decisions (as in presumed consent to organ donation), or the
physical environment for decision-making may be designed to apply influence (as
where the office smoking zone is placed at a distance from the work area). Both of
these examples nudge individuals’ decisions by relying on human inertia.
The Second Degree nudge involves a greater impact on individual autonomy than
the First Degree nudge since the targeted individual’s behavioural or volitional
limitations and ‘automatic’ responses will in practice lead him or her to accept the
nudge with limited awareness and reflection. It is, nevertheless, the case both that
such a nudge will shape an individual’s decision and that the target of the nudge
would be capable, on reflection, of realising that a nudge has been administered and
assessing its broad effect.
A ‘Third Degree nudge’ offers a yet more serious intrusion on autonomy because
it involves behavioural manipulation to an extent that other nudges do not. Thus,
framing devices can be used to shape the decisions and preferences of an individual
in a manner that is resistant to unpacking in so far as assessing the nature and extent
of the nudge is not readily achieved by reflection.22
Thus, a campaign to promote
healthy eating might focus on certain negative consequences that are associated with
20
See Hausman and Welch, n 14 above, 127. 21
See e.g. Behavioural Insights Team, Behaviour Change and Energy Use (London: Cabinet
Office, 2011) Chapter 2. 22
On preferences being sensitive to the way a choice problem is described or ‘framed’ see: P.
Slovic, ‘The Construction of Preference’ (1995) 50(5) American Psychologist 364 - 371; A.
Tversky and D. Kahneman, ‘The Framing of Decisions and the Psychology of Choice’ (1981)
211 Science 453-458; J. Payne, J. Bettman and E. Johnson, ‘Behavioural Decision Research:
A Constructive Processing Perspective’ (1992) 43 Annual Review of Psychology 87 – 131.
6
poor diets so as to structure decisions by giving dominance to particular
considerations (‘Don’t lose your looks, junk the junk food!’). Reflection makes the
target aware of the emphasis on ‘looks’ in the example given but not of the extent of
preference adjustment involved in expressing the eating decision as one involving a
loss, rather than gain; in giving ‘looks’ a dominant position in decisions about healthy
eating; and in applying the emotive power associated with a loss of looks. The result,
in cases of preference shaping, is that even reflection makes little difference to the
individual – it only reinforces the pursuit of the ‘shaped preferences’. In such cases,
the Third Degree nudge does not so much push the tiller as rig the compass bearing.
Another form of Third degree nudge will often rely on salience and affect to
produce its result. Typically, there is use of the behavioural economist or
psychologist’s finding that people are influenced by novel, personally relevant or
vivid examples of, say, a harm.23
In such instances the message receiver will be
motivated emotionally in a manner that stands in the way of reflection. Thus, in the
smoking context, the cigarette pack would show a graphic display of a corpse.
(Alternatively, a smoker could be shown a video of a loved one having a heart attack.)
The element of ‘manipulation’ stems from the use of a level of emotional power so
that this: ‘Blocks the consideration of all options and threatens the agent’s ability to
act in accordance with her or his own preferences (as opposed to someone else’s).’24
The difference between a Second and a Third Degree nudge is that, with the
former, the message receiver has the practical potential to uncover the nudge, and
assess its extent, by the exercise of reflection whereas, in the latter, there is a material
or complete25
blocking of such reflection, or the neutralising of reflection. The
distinction between Second and Third degree nudges does not turn on a sharp
differentiation of cognitive, volitional and emotional functions and manipulations, nor
is it to presuppose a lack of linkage between these functions – the literature of social
psychologists and sociologists of emotion offers an extensive discussion of the extent
and nature of such links26
– it is suggested here that Third Degree nudges are
distinctive in so far as decision framing through emotional or any other form of
influence (or combination of influences) is sufficient materially to obstruct reflection
or the assessing, on reflection, of the extent, nature and degree of the nudge. 27
23
J. Blumenthal-Barby and H. Burroughs, ‘Seeking Better Health Care outcomes: The Ethics
of Using the “Nudge”’ (2012) 12(2) Am. J. of Bioethics 1-10, 4. 24
ibid. 5. 25
An example of complete blocking would be the use of subliminal messaging – which
imposes the nudge without the nudged individual’s awareness at the time or even afterwards.
This would be unannounced, unpublicised subliminal messaging - though a government
might openly declare that it has a policy of using subliminal messaging to control e.g. a health
issue. Thaler and Sunstein argue that they endorse Rawls’ publicity principle as a limitation of
nudging and that subliminal messaging is, accordingly, not acceptable because it lacks
transparency, see: Thaler and Sunstein 2008, n 2 above, 243-246 and the discussion below. 26
See e.g. J. Storbeck and G. Clore, ‘On the interdependence of cognition and emotion’
(2007) 21 Cognition and Emotion 1212-1237; C. Izard, J. Kagan, and R. Zajonc (eds)
Emotions, Cognition, and Behavior (Cambridge: CUP, 1984); T. Kemper (ed) Research
Agendas in the Sociology of Emotions (1990); T. Dalgleish and M. Power (eds), Handbook of
Cognition and Emotion (Chichester: Wiley, 1999); T. Dalgleish and M. Power, Cognition
and Emotion (Hove: Psychology Press, 2008). 27
The difference between Second and Third degree nudges may thus be a difference of
degree (no pun) rather than kind. Second degree nudges may have an emotional dimension
and they may also structure choices - the issue is whether the nudge materially impedes
reflection or an awareness, through reflection, of the extent, nature and degree of the nudge.
7
Table 1: Three Degrees of Nudge
Typical
Characteristics
Example Impact on
autonomy
First Degree Nudge
Supply of simple
information or a
reminder with the
aim of improving
the target’s
capacity to make an
informed, rational
and conscious
choice.
1. Health warning
on cigarette pack.
2. Reminder to fill
in tax return.
Respects the
autonomy of
decision-maker and
enhances target’s
rationality.
Second Degree
Nudge
Behavioural or
volitional
limitations are
exploited so as to
bias decisions in a
favoured direction.
1. An opt-out organ
donor regime is
instituted.
2. The office
smoking zone is
placed at a distance
from the work area.
The target could,
on reflection,
unearth the nature
and effect of the
nudge – but is
unlikely to do so
because of
behavioural
limitations and
their tendency to
exhibit an
‘automatic’
response.
Third Degree
Nudge
Framing strategies,
emotional
responses or covert
techniques are used
to influence
decisions or shape
preferences.
1. A campaign
promotes healthy
eating with the
slogan: ‘Don’t lose
your looks, junk
the junk food!’
2. Shocking images
are used to control
behaviour – as
when photographs
of lung cancer
victims are used to
control smoking.
3. Unpublicised
subliminal TV
messages are used
to encourage e.g.
healthy eating or
abstention from
smoking.
The target is
influenced but
reflection is
obstructed or
reflection
materially fails to
unpack the nature
and extent of the
decision or
preference shaping.
8
The distinctions between these three degrees of nudge are important in assessing both
the potential effectiveness and the acceptability of nudging by different methods in
different contexts. (This is the case in spite of our ability to conjure up instances at the
borderlines of these categories.) Questions of acceptability will be considered below
but, at this stage, it suffices to note that some degrees of nudge present much more
serious representational and ethical issues than others. It follows that there are real
dangers in adopting a loose conception of nudge. The most serious of these is that the
most benignly uncontentious examples of nudge may be highlighted in debates and
that this impedes critical reflection on the more controversial aspects and versions of
the device.
LIMITS TO THE EFFECTIVENESS OF NUDGE
Nudges are seen, by many politicians, to be an alternative to ‘command’ regulation28
and it is appropriate, accordingly, to measure the nudge against the yardsticks
commonly used in assessing regulatory strategies.29
Relevant issues are whether
nudge produces properly targeted outcomes effectively and whether nudging serves
representative and ethical values acceptably.
On the question of effectiveness, the nudge has practical limitations – some of
which are general and some of which apply especially to certain degrees of nudge.
The first limitation to be noted is one that applies to all degrees of nudge and stems
from nudge’s strategy of aiming to achieve results by focussing on the decision-
making of the individual. It has been pointed out that such a focus may fail to address
the causes of, or provide the solution to, a number of problems.30
Obesity, for
example, may be seen by nudgers as the result of an accumulation of poor individual
decisions (which can be improved by nudges) but this vision fails to take on board a
host of biological, social and cultural causes of obesity. It has been argued, for
instance, that the rise in obesity of recent years is not to be explained in terms of
personal responsibility but in: ‘..a toxic combination of readily available cheap high
energy food and drink, fewer opportunities for manual labour, an increase in car
ownership, changing social norms concerning cooking and eating, and other features
of the “obesogenic” environment.’31
If personal consumer decisions are only one
element in the causation of obesity, nudges of those decisions are likely to prove of
limited effect in dealing with the essential problem. It is likely that obesity has to be
28
See the evidence of the Minister for Government Policy, Oliver Letwin, to the House of
Lords Select Committee, 2011, Q. 703, reported at para. 5.3. 29
On yardsticks for measuring regulation see generally R. Baldwin, M. Cave and M. Lodge,
Understanding Regulation (Second Edition) (Oxford: OUP, 2012) Chapter 3. 30
See: J-F Menard, ‘A “Nudge” for Public Health Ethics: Libertarian Paternalism as a
Framework for Ethical Analysis of Public Health Interventions?’ (2010) 3 Public Health
Ethics 229-238. 31
Evidence of Professor Thomas Baldwin, House of Lords Select Committee, 2011, para. 7.4.
Other witnesses echoed this point: see the evidence of Dr Ian Campbell, Medical Director,
Weight Concern (noted at para.7.8). See also: Foresight and BIS, Tackling Obesity: Future
Choices (2007). For a discussion of behaviour change interventions and obesity see Appendix
4 of House of Lords Select Committee, 2011. For similar arguments in relation to reducing
car use see paras. 7.26 –7.48 and Appendix 5.
9
addressed with a combination of personal, normative, cultural and environmental
measures (regulatory and non-regulatory) rather than left to nudging alone.32
Nudges that are aimed at individuals, moreover, will not always prove effective
when the undesirable behaviour at issue is the product of collective processes and
policies (for example where competing interests and pressures within an enterprise
produce a corporate decision that harms the environment).33
Individually - targeted
nudges are likely to impact on collectively-generated decisions only when the latter
are sensitive to changes in the targeted individuals’ decisions (as where consumers are
given information on corporate environmental records, this affects sales of the
corporation’s products, and the corporation is sufficiently responsive to amend its
environmental strategies as a result). It follows that, if many modern-day problems are
organizationally generated, governmental inclinations to avoid the ‘nanny state’ by
relying on ‘non-regulatory’ controls and individual responsibilities may hit the mark
only under certain conditions.
A second limitation of effectiveness can be expected to vary across degrees of
nudge. It is that individuals’ responses to nudges will differ across divergent
institutional, social, economic and cultural contexts.34
A nudge to encourage recycling
may not work as well in a disadvantaged community as in an affluent district.35
Nor
will automatic enrolment into a retirement plan (with opt-out) operate to increase
savings in the body of the population that does not earn enough to be able to put this
money aside.36
Nudges, and different degrees of nudge, accordingly, can be expected to echo the
tendency of traditional regulatory styles to vary in their effectiveness when targeted at
different types of regulated concern and at people with different characteristics. Table
2 below looks at four examples of nudges as applied to four different models of
individual. It indicates whether the effect of a given nudge on the particular target is
likely to be high, medium or low. The objective of the discussion here is not to give
an exhaustive (and exhausting) account of variations in effectiveness across the huge
number of possible ways to implement different nudge tools. The aim is to illustrate
the kinds of variation in effectiveness that can be expected across not only degrees of
nudge but across different nudge tools (even tools of the same degree). The four
nudges set out in Table 2 are thus chosen to exemplify different degrees of nudge but
also different nudge tools of the same degree.
The breakdown of individuals used in Table 2 is one familiar in the regulation
literature37
and refers to those characterised respectively as: well-intentioned and high
capacity; well-intentioned and low capacity; ill-intentioned and high capacity; ill-
32
House of Lords Select Committee, 2011, paras. 7.6 and 8.14 and evidence of Dr M.
Hillsdon. 33
In contrast are instances in which the relevant decisions are made by individuals and those
decisions are nudged. In such cases the nudging of individuals can produce reactions on the
part of corporations – as where citizens are nudged to smoke fewer cigarettes and corporate
producers adapt their marketing strategies as a result. 34
See: e.g. L. Bovens, ‘Nudges and Cultural Variance’ (2010) 23 Know.Tech. Pol. 483-486. 35
John, Smith and Stoker, n 14 above, 369. 36
Menard, n 30 above, 235. 37
See e.g. J. Black and R. Baldwin, ‘When Risk-Based Regulation Aims Low: A Strategic
Framework’ (2012) 6 Regulation and Governance 131-148; R. Baldwin, M. Cave and M.
Lodge, Understanding Regulation 2nd
ed. (Oxford , 2012) pp. 230-231; R. Kagan and J.
Scholz, ‘The Criminology of the Corporation’ in K. Hawkins and J. Thomas (eds), Enforcing
Regulation (Boston: Kluwer-Nijhoff, 1984).
10
intentioned and low capacity. ‘Intentioned’ here refers to the extent to which the
targeted person espouses the same objective as the nudger. Thus, a well-intentioned
target will be interested in, and strongly disposed in favour of the behaviour that the
nudger sees as virtuous. Ill-intentioned targets will not espouse such objectives and
may even be actively opposed to behaving in the manner that the nudger sees as
virtuous.
‘Capacity’ refers to the ability of that person to gain, receive absorb and act on
information. A high capacity individual will thus be able to receive messages (even
complex ones) effectively and will be able to adjust his or her behaviour in response
to such messages. Whether such an individual will be inclined to adjust their
behaviour in accordance with the message received will, of course, depend on their
intention rather than capacity. High capacity individuals who are not well-intentioned
and not inclined to act in accordance with a message will be very able to adjust their
behaviour so as to reject that message and act in ways that are inconsistent with it.
Such individuals will possess a high ability to ‘unearth’ nudges, such as defaults, and
to resist these. Low capacity individuals will struggle to absorb and act on even
simple messages, even when disposed so to act. Low capacity individuals who are ill-
intentioned, will, moreover, have very limited ability to adjust their behaviour so as to
reject messages that they disagree with and to act in ways that are inconsistent with
such messages. They will, in turn, possess poor abilities to ‘unearth’ nudges such as
defaults, and resist these.
Table 2: The Anticipated Impact of Nudges on Different Targets.
Information
Mechanism:
Simple
nutritional
information is
provided on
food packaging.
(A First Degree
Nudge)
Default:
An opt-out
regime is
adopted for
organ
donation.
(A Second
Degree Nudge)
Design:
The smoking
zone is placed at
a distance from
the office work
area.
(A Second
Degree Nudge)
Warning:
Shocking
photographs of
lung cancer
victims are used
to control
smoking.
(A Third Degree
Nudge)
Well-intentioned and
High Capacity
High High High High
Well-intentioned and
Low Capacity
Low High High High
Ill-intentioned and
Low Capacity
Low High Medium High
Ill-intentioned and
High Capacity
Low Low Low Medium
11
Empirical evidence on the above indications is, as yet, thin on the ground (a matter to
be returned to below) but, if there is some truth in the estimations summarised in
Table 2, this offers at least three relevant messages for potential nudgers. First,
different degrees of nudge, different nudge tools, and different ways of implementing
given tools can be expected to bring different effectiveness profiles. Thus, the
successful use of a First Degree nudge, such as the provision of simple information,
can be expected to be far more dependent on the capacity of the target than a Second
Degree nudge such as a design, or a default approach or a Third Degree nudge such as
a shock image. The First Degree nudge, after all, depends on stimulating high quality
conscious decision-making. With targets of low capacity it can be expected that
Second and Third Degree nudges will bring special potential – though it can also be
seen in Table 2 that even across different tools of the same degree there may be
variations in effectiveness with regard to low capacity targets.
A second message is that even a First Degree nudge, such as simple information
supply, may prove effective only with a small minority of targets. Third, it can be
anticipated that targets who are ill-intentioned and high capacity will tend to be
difficult to nudge with a high degree of effectiveness whatever degree or tool of
nudge is employed. 38
At this point it should be emphasised that the above discussion does not constitute
an argument against nudging per se but it offers a caution that the respective
limitations of different degrees, tools and modes of implementing nudges need to be
borne in mind if the nudge approach is to be deployed astutely.
A further limitation on the effectiveness of nudging is presented by the possibility
of the counter-nudge, and here there is a contrast with traditional command
approaches. If legislators ban the sale of alcohol above a certain strength, they deal
with the issue definitively. If they rely on a nudge to control this activity (e.g.
imposing a requirement that a person registers before purchasing such alcohol) there
is the potential for interested parties to ‘counter-nudge’.39
Thus, the supermarkets
have the chance to respond by stacking high strength alcohol next to the till or by
advertising it more intensively. The marketing and advertising sectors of the private
sector may, indeed, prove formidable opponents for the governmental players of the
nudge game. It is unlikely, for instance, that nudgers who seek to control obesity will
ever match the power to influence that is provided by the promotional budgets of the
fast food industry. This is a point that applies across the range of nudge categories,
though the higher the degree of nudge used, the more severe will be the response
needed to contest this.
When nudging operates over time, there is another issue that goes to the
effectiveness of the device. Nudges that are genuinely paternalistic (Second and Third
Degree nudges) may undermine relationships that are based on trust when the practice
38
Table 2 suggests that, even with the default and design strategies that are discussed, targets
of ill-intention and high capacity may use their cognitive resources to unmask the nudges and
circumvent these. See also the discussion of the limits of regulating internet users’ behaviour
through ‘code’ in A. Murray, ‘Internet Regulation’ in D. Levi-Faur (ed), Handbook on the
Politics of Regulation (Cheltenham: Edward Elgar, 2011). 39
It is conceded that the targets of legislation may respond by strategies such as creative
compliance or breaching the law – what they cannot do is counter-legislate.
12
of nudging is uncovered. Thus, in the field of medicine it has been warned that nudges
can be seen as scare tactics by patients and that this can alienate them from their
doctors. The use of information selectively or emotively can thus act to the detriment
of treatment in the longer term.40
Similarly, it has been argued that nudges may lead
patients to see their physicians as operating a ‘policing’ function and that this may
lead the former to withhold important medical information from those who treat
them.41
The problem will be all the greater if the nudger is seen to be manipulating the
truth in ‘constructing a narrative’ for the purposes of nudging – as, for example,
through an inaccurate or ‘aspirational’ representation of general practice such as:
‘Nearly all patients consent to this procedure.’
As for nudge’s effectiveness as a device for avoiding need for legal rules, this is
easy to exaggerate. Governments can nudge individuals directly but they will often
use businesses as their nudging agents for all degrees of nudge. This occurs, for
example, when retailers are legally required to provide consumers with certain
messages or items of information. In such regimes it is true that consumers do escape
from the clutches of command and control laws but this is by no means the case for
the affected businesses.
As far as producing properly targeted outcomes effectively, it cannot be assumed
that nudges (of whatever degree) are any less prone to the triggering of unintended
consequences than other forms of regulatory or non-regulatory intervention.42
Thus, it
has been pointed out that consumers may override intuitive responses or react
perversely to nudges.43
Food that is labelled as healthy may lead to excessive
consumption.44
Positive information on low levels of a certain unhealthy ingredient
may blind the consumer to high quantities of another unhealthy component,45
and a
positive message about one aspect of a food may induce underestimations of other,
harmful, qualities.46
A further targeting challenge arises when opinions differ or develop on what such
outcomes involve – and these challenges may differ across degrees of nudge. First
Degree nudges encourage conscious deliberation on the ‘think’ model and operate
with a mechanism that can cope with differences and changes in perceptions
regarding desired outcomes (e.g. new ideas about the right balance between alcohol
and carbohydrate consumption). The danger with nudges that influence automatic
responses (Second and Third Degree nudges) is that it is the paternalistic nudger that
defines desired outcomes and this may lead to a false sense of security concerning the
desirability of those outcomes.47
The problem is that, even if the nudger’s views on
40
Blumenthal-Barby and H. Burroughs, n 22 above, 5. 41
ibid. 2. 42
An assumption apparently made by the Minister, Oliver Letwin MP, in giving evidence –
see House of Lords Select Committee, 2011, para. 5.7. 43
See O. Amir and O. Lobel, ‘Stumble, Predict, Nudge’ (2008) 108 Col. L. Rev. 2098-2137,
2116. Ill-intentioned, high capacity targets may be especially prone to perverse reaction. 44
House of Lords Select Committee, 2011, para. 7.11. 45
An investigation by The Daily Telegraph in March 2014 revealed that foods promoted as
‘low fat’ and ‘low calorie’ often had more harmful effects on heath than their full fat
equivalents. The campaign group Action on Sugar protested that manufacturers were
‘misleading’ shoppers. See http://www.telegraph.co.uk/health/healthnews/10668189/Low-
fat-foods-stuffed-with-harmful-levels-of-sugar.html (Last visited 8.3.14). 46
T. Marteau et al ‘Judging Nudging: Can Nudging Improve Population Health?’ (2011) 342
BMJ 263, 264. 47
John, Smith and Stoker, (2009), n 14 above, 368.
13
outcomes coincided initially with those of the consumers or citizens being protected,
the two strands of thought may, over time, diverge because such nudges are based on
paternalism rather than the fostering of an ongoing dialogue about objectives. In such
cases, nudge may produce results other that those that the nudged parties would have
sought.
Finally, it should be noted that numbers of commentators have stressed that there
has emerged very little concrete evidence on the effectiveness of nudge strategies. To
an extent this may stem from the paucity of studies that deal with the performance of
nudges.48
The House of Lords Science and Technology Committee, indeed, stated in
2011 that it had been provided with no examples of significant change in the
behaviour of a population having been achieved by non-regulatory measures alone.49
Its report concluded that non-regulatory measures were often not likely to be effective
when used in isolation and that, usually, a range of policy tools, of both kinds, would
be need to bring about change effectively.50
The Committee stressed that more
rigorous steps should be taken within government to ensure the effective evaluation of
behaviour change interventions.51
REPRESENTATIONAL AND ETHICAL ISSUES
Nudge, with its philosophy of ‘libertarian paternalism,’ gives rise to a number of
representational and ethical concerns. A first of these is that many decisions to nudge
lack the transparency and public consideration that are normally associated with
command regimes.52
If a government issues a law that prohibits citizens from
smoking in public places, this is a mode of control that is open, discussed, and
implemented after representative processes have been followed.53
If nudging is used,
the process used to effect a nudge may be far more secretive—the nudge, for instance,
may flow from an administrator’s decision on the design of a public building: a
decision not subjected to advanced disclosure or debate. As Luc Bovens has said,
nudges ‘typically work better in the dark’.54
This may be seen as relatively
unobjectionable in the case of First Degree nudges, since these contribute positively
48
Marteau, n 26 above, 264. 49
House of Lords Select Committee, 2011, para. 5.6. 50
ibid. paras. 5.13; 8.14. 51
ibid. Chapter 6 and paras. 8.21-8.22. Others have suggested that, given the variety of
nudges, such work should deal with a series of issues: What works? For whom? In what
circumstances? For how long? How effective would nudges be compared to other
interventions? See e.g. Marteau and colleagues at Marteau (2011) n 26 above , 264. 52
See R. Brownsword, ‘Code, Control, and Choice: Why East is East and West is West’
(2005) 25 Legal Studies 1-21 who contrasts the transparency of ‘command’ (‘East coast’) and
‘design’ (‘West coast’) controls. See similarly: L. Tien, ‘Architectural Regulation and the
Evolution of Social Norms’ (2004) 9 International Journal of Communications Law and
Policy 1. On control through ‘design’, ‘code’ or architecture’ more generally see L. Lessig,
Code and Other Laws of Cyberspace (New York: Basic Books, 1999); A. Murray and C.
Scott, ‘Controlling the New Media: Hybrid Responses to New Forms of Power’ (2002) 65
MLR 491. 53
The transparency of ‘command’ approaches should not, however, be exaggerated – many
commands are promulgated through delegated legislation and other varieties of soft law. See
generally R. Baldwin Rules and Government (Oxford: OUP, 1995). It is arguable,
nevertheless, that nudges can be instituted by methods that are far less transparent than even
the processes used to introduce soft-laws. 54
See Bovens, n 14 above, 3 ( referring to nudges that would be Second Degree nudges in
the framework proposed by this article).
14
to conscious individual decision-making, but a lack of transparency regarding Second
and Third Degree nudges would be far more contentious.
It is true that, where legislation is used to institute the nudge – as when a new law
demands that supermarkets place fruit next to the till – there will be the normal public
debate that attaches to the legislative process. Many nudges, however, will not
demand legislation and may be triggered administratively, as may, for instance:
informational nudges, administrative default rules, publicity campaigns, alterations in
the physical environment and the messaging of social norms.
The proponents of nudge might protest that nudges of all degrees can be disclosed.
Thaler and Sunstein, indeed, state that they object to nudging by subliminal
messaging because they endorse Rawls’ publicity principle as a limitation of nudging
and subliminal messaging is, accordingly, not acceptable because it lacks
transparency.55
The problem, however, is that, although it is possible for a
government to state openly that it is using Second or Third Degree nudges in an area,
there are reasons to be pessimistic about the prospects of this ocurring so that this
enhances rational decision-making or transparency. First, there is the Bovens point,
noted above – if the students are told that the cafeteria food is arranged in order to
push them towards a healthy diet, this is likely to undermine the effectiveness of the
nudge.56
Second, a general disclosure about nudging (‘We have designed this working
environment to help you make healthy decisions’) may prove unhelpful in rendering
the nudge’s existence or nature evident at the time when a particular decision is
taken.57
After all, the critics would say, nudges (of the Second Degree and Third
Degree) are applied on occasions that allow the exploitation of an individual’s
weaknesses of willpower, emotion and rationality and this is not a context likely to
encourage advertence to the fact or extent of a nudge. Showing the drowning bather a
flipchart about learning to swim may be poor timing.
In further defence of the nudge it might be suggested that there is little need for
debate when all the nudger is doing is making targets’ lives ‘longer, healthier and
better’ and trying to ‘influence choices in a way that will make choosers better off as
judged by themselves’.58
The nudge, on this view, conduces to the decision that the
target would have arrived at had he or she paid full attention and possessed complete
information, unlimited cognitive abilities, and complete self-control.59
The limitation
of this argument is that Second and Third Degree nudges do not involve seeking to
identify the conception of welfare that the target would espouse.60
In contrast, when
nudging goes beyond the supply of simple information, or reminders, at the ‘First
Degree’ level (which can be said to enhance the target’s conscious decision-making)
what is involved is the nudger acting in a semi-covert or covert manner to further the
nudger’s own conception of the target’s welfare – which may involve re-shaping the
target’s idea of their own welfare.
55
Thaler and Sunstein 2008, n 2 above, 243-246. 56
See Bovens, n 14 above, 3-4. 57
Brownsword, n 52 above, 14 - 17 makes the additional point that ‘design’ approaches can
‘embed’ solutions so that those affected in the future have little consciousness of prior design
decisions and no input into the ways in which they are controlled. 58
See Thaler and Sunstein 2008, n 2 above, 5. 59
ibid.6. 60
On the argument that targets’ desires and preferences tend to be shaped inevitably and that
this justifies nudging because there is usually no autonomy to be interfered with, see the text
following footnote 72 below.
15
This would not be such a great worry if there was some objective welfare to be
pursued without the possibility of contention. Such a scenario, however, may be less
realisable than first appears. Thus, a nudge might aim to further ‘better health’ on the
assumption that this is uncontroversial but, as Gregory Mitchell has pointed out:
‘Many may agree in the abstract that better health is preferable to worse health, but
when the choice is framed as enjoying life-shortening but intensely pleasurable vices
during one’s college days versus abstaining during college to gain a couple of extra
boring years at an advanced age, then better health may not look quite as good.’61
The danger is that the enthusiastic proponents of nudging62
will be slow to
anticipate conceptions of welfare or preferences that they see as non-sensible or which
stem from perspectives that they do not share. They may, moreover, fail to anticipate
that nudges may prove contentious not merely because they seek to further contested
visions of welfare but because the costs of furthering those visions, or the
distributions of such costs, or the costs of opting out of nudges may be controversial
matters in themselves. A further concern of nudge’s critics may be that the banner of
libertarian paternalism may be used as a cover for the pursuit of social objectives
(such as lowering hospital’s administration costs) rather than the welfares of the
nudged individuals.63
Proponents of nudge, of course, have a fall-back defence against the accusations
of illiberalism that apply to Second and Third Degree nudges: they stress, as noted,
that a nudge should be ‘easy and cheap to avoid’.64
A difficulty here is that there is an
inherent tension in the nudging argument that goes to the feasibility of opt-out. The
nudge has power in so far as it impacts on the decision-maker of limited cognitive
capacity, information and self-control. Those very limitations, however, mean that the
target is unlikely to be well-placed to exercise the opt-out in cases of Second Degree
nudges. With Third Degree nudges, the position is worse because the target will be
‘blocked’ from resorting to the opt-out. When a nudge is used to shape the target’s
preferences (as when a vivid example of a harm is used to trigger aversion) the opt-
out is taken out of play. As Mitchell argues, more generally: ‘If choice is as ‘sticky’as
Thaler and Sunstein claim, then people who lack the means to determine their true
preferences in a given setting cannot make any real use of the opt-out provision and
will simply stick with the default set by the libertarian paternalist. For these people
libertarian paternalism is just paternalism.’65
For such reasons, nudges have been said
to be, on their face, as threatening to liberty, broadly understood, as is overt
coercion.66
The opt-out defence is also vulnerable to the criticism that it not only assumes a
level of competence, rationality and volitional control that contradicts the
underpinning assumptions of behavioural economics, but it understates the extent to
which opt-outs discriminate against parties who are less able to exercise them. An
opt-out that is a mere mouse click is easy enough to most persons but not to the
person who does not possess a computer. An organ donation opt-out that is recorded
61
See G. Mitchell, ‘Libertarian Paternalism is an Oxymoron’ (2005) 99 N. Western U. Law
Review 1245, 1268. 62
On nudge as ‘never knowingly undersold’ see R. Baldwin, The New Scholarship:
Celebrating the ‘I’ in Ideas LSE Law, Society and Economy Working Papers 5/2012.
Available at: http://www.lse.ac.uk/collections/law/wps/WPS2012-05_Baldwin.pdf 63
Blumenthal-Barby and H. Burroughs, n 22 above, 4. 64
Thaler and Sunstein, 2008, n 2 above, 6. 65
Mitchell, n 61 above, 1254. 66
See Hausman and Welch, n 14 above, 130.
16
on a driving licence would not be very accessible to a committed cyclist, and some
smokers with mobility issues may not think that the 100 metre journey to the office
smoking zone is an easy opt out from the design nudge of the departmental architect.
In the medical field, it has been pointed out that written opt-outs are not that useful to
linguistic minorities, and illiterate persons.67
A further concern is that, even if exercising one opt-out may be relatively easy,
cumulations of nudges (of the Second and Third Degree68
) produce an aggregate
burden that will weigh down persons who would seek to negotiate their own way
through everyday decision shapings so as to pursue their own aims and objectives. In
a world of nudge, the worry is, the individual fights an unending battle to construct
his or her own preferences and to play the opt-out game in order to further those
preferences. Their autonomy, they may feel, dies the death of a thousand nudges.
To construct a simple example, let us suppose that supermarkets agree with the
Government that on-line shoppers will be presented with default purchases that
encourage healthy lifestyles and that the retailers in question nudge with enthusiasm.
The shopper may avoid the default acqusition of fresh apples on supermarket X’s
website by a mouse click but then has to click to avoid purchasing broccoli, cod liver
oil tablets, aspirins, glucosamate pills and thirty other products. This arrangement
raises a number of issues. In the first instance, there is a question of how a nudge is
conceived for the purposes of opt-out. The ‘apples nudge’ is avoided by a single click
but the ‘healthy lifestyle’ nudge requires over thirty clicks to neutralise – which is
likely to prove wearisome if the process afflicts all on-line shopping episodes. On one
view, each nudge is avoided easily and complies with Thaler and Sunstein’s
conditions for nudging. On another view, the cumulative burden of opting out of this
package of defaults is unacceptable.
A further worry about Second and Third degree nudging is prompted by the
example under consideration. Where a Government uses agents, such as retailers, to
implement such nudges this creates a specific set of problems.69
Not only are there
issues of consistency of interpretation, transparency and accountability, but there are
challenges to the opt-out. Where different supermarkets implement the nudge in
different ways, the consumer has to pay special attention to each opt-out because
some nudgers’ approaches and some ‘nudged’ products will be more questionable
than others. (Will the dosages of glucosamine sulphate sold by supermarket Y’s
default actually bring benefits that justify the costs? Is the cardiac case for a daily
aspirin made out?) For consumers, the effort required to address these issues only
adds to the burdens of opting-out. They may resent having to negotiate their way
through a brand-specific lifestyle training session every time they order some
groceries.
Nor can it be assumed that the consequences of a failure to opt-out will be
uniform across populations. Thus, for example, presumed consents for HIV testing
may prove especially threatening to individuals who are liable to a range of
particularly negative consequences if they fail the test – such as domestic violence or
67
See M-A Jacob, ‘Another look at the presumed-versus-informed consent dichotomy in
post-mortem organ procurement.’ (2006) 20(6) Bioethics 293-300. 68
This point applies to a lesser extent to First Degree nudges since even warnings and
simple items of information may cumulate to a mass that is seen as oppressive by the targets. 69
Agency issues, of course, add to dangers that nudges will cause unintended consequences
and this further undermines the argument that nudges are more easily targeted than
commands.
17
job loss.70
Where there are such divergencies of consequence, and these are not taken
on board, the danger is that calculations about the balance between the value of the
nudge and the costs of opt-out will produce nudges that discriminate against
vulnerable parties.
A broader issue of fairness also arises with Second and Third Degree nudges –
they impose costs on sensible persons in order to enhance the welfare of others who
behave irresponsibly. To take an example, let us suppose that a government is
concerned to control excessive alcohol consumption. It passes a law to require all
retail outlets to place alcohol in locked cabinets, out of sight of shoppers. The (Second
Degree) nudge derives from the conscious effort that is required in order for the
shopper to obtain the alcohol (by specific request for service) and from the
neutralising of any sales-enhancing layouts by the retailer. Leaving aside whether this
nudge would limit consumption effectively, it can be noted that the nudge involves an
inconvenience for, amongst others, the 93 per cent of the mature male, and 96 per cent
of the mature female, population who can limit their consumption to below ‘harmful’
levels when left to their own devices and exposed to the nudges of retailers, brewers
and distillers. 71
There is, accordingly an issue of distributional justice here and some
parties will object to this redistribution from the rational/responsible to the
irrational/irresponsible.72
A second defence of nudging is of special relevance to Second and Third Degree
nudges and argues that there is always a nudge from some direction so that peoples’
preferences and choices are inevitably structured and influenced (either intentionally
or unintentionally) and that, therefore, it is best that the nudge is applied in a welfare
enhancing direction.73
Thus, it might be contended that, uncontrolled, the astute
supermarket would be likely to place a profit maximising product next to the till, or
the less organised supermarket might rotate products randomly on the till display. Far
better, it would be said, if the supermarket was required to place the fruit next to the
till in order to nudge consumers towards healthy options.
A first objection to this defence is that it is wrong to assume that all preferences
and choices are framed or presented in unacceptable ways. In many decision-making
contexts, the decision-maker is capable of identifying his or her own preferences and
is faced with information that he or she is quite capable of processing and acting on so
as to serve their interests and preferences.74
The real issue, on this view, is not
whether the decision is shaped but whether the pre-nudge decision-making context is
seen as manageable and acceptable. Even when nudged by, say, a retailer, it is routine
for the consumer to be content with the rationality and level of autonomy that he or
she exercises. The decision environment may not be ‘true’ but it is seen as acceptably
manageable. (The consumer exercises his or her autonomy here in accepting that they,
rather than someone else, will deal with the decision-making context in all its
imperfections.) Thus, many vegetable purchasers know what they are looking for and
70
C. Hanssens, ‘Legal and Ethical Implications of Opt-out HIV Testing.’ (2007) 45 Clinical
Infectious Diseases 232-239. 71
In the 45-64 year old grouping, the evidence is that 7% of men and of 4% of women drink
to a ‘harmful’ level (over 50 units per week for men and 35 units for women) – see Institute
of Alcohol Studies, Alcohol Consumption Factsheet (IAS, 2013) 11. 72
See Mitchell, n 61 above, 1270. 73
First Degree nudges arguably do not demand that this defence is made since they enhance
conscious decision-making. 74
See the concession of this point at Thaler and Sunstein 2008, n 2 above, 10.
18
expect the greengrocer to place the best produce at the front of the stall and to serve
from the back.
The advocates of Second and Third degree nudging, it can thus be argued,
interfere with the autonomy of individuals too readily and nudge theory can be said
not yet to have produced either a way of separating out unacceptable from acceptable
choice structurings or of identifying those circumstances in which decision-making
autonomy is so prejudiced by shaping influences that this justifies intervention
through another nudge. 75
A closely related line of nudge-defence should be considered at this point. This
contends that it is wrong to object to Second and Third degree nudging as an attack on
individual autonomy because such an objection presupposes an idealised version of
‘true’ autonomy that is based on an assumption of fully-informed, rational and
unshaped decision-making. The reality, the nudge proponents argue, is that the
‘autonomous choices’ that individuals generally make are usually shaped by a host of
features of the consumption environment, they are often the products of cognitive and
volitional limitations, and, as a result, there is rarely an exercise of unimpaired and
‘true’ autonomy. The counter to this argument, however, is not merely the point just
made (that some choice framings are seen as acceptable) it is also that the pro-nudgers
are too quick to portray some preferences as irrationalities. Thus, some of the ‘biases
and blunders’ that Thaler and Sunstein cite as causes of poor decisions76
can be said
to be preferences that deserve to be respected rather than cognitive or volitional
failings that need to be reacted to with a nudge. Loss aversion, for instance, is said by
Thaler and Sunstein to be a ‘matter that people can get pretty emotional about’77
so
that they value not losing things far more than gaining the same things. This produces
inertia and leads to decisions that do not serve peoples’ own interests.78
It is a
‘fallibility’ that the two authors suggest can be corrected with a nudge. The nudge-
sceptics, however, would argue that this is not a case of impaired autonomy but of a
choice that the nudgers disagree with. The sceptics might add that, if a person has a
preference for the status quo, it is their life, their inertia, their choice, not that of the
nudger.
To return to reservations about the ‘nudges are unavoidable’ argument, a second
objection is that, even if the premise is assumed to be correct, the conclusion does not
follow. Thus, it can be argued that the supermarkets’ propensities to nudge consumers
in unhealthy directions (leading, for example, to the consumption of excessive
amounts of sugar) do not necessarily demand that the state responds to such nudging
with another nudge (or, indeed, with a nudge of a Second or Third degree as opposed
to a First Degree). In many contexts, it could be argued, it would be more effective,
more accountable and more transparent to react with, say, a command approach and
revise the laws on food ingredients (e.g. to limit the maximum concentrations of sugar
that are deemed safe to sell).
There is, indeed, an argument that to respond to nudges with state-applied counter
nudges of the Second and Third Degree is exactly the wrong policy in those cases
where it causes citizens and consumers to be exposed to layers of competing nudges.
75
See Mitchell, n 61 above, 1249-50. Thaler and Sunstein do, however, offer a discussion of
the circumstances in which people are most, and least, likely to make good decisions – see
Thaler and Sunstein 2008, n 2 above, Chapter 4. 76
Thaler and Sunstein 2008, n 2 above, Chapter 1. 77
It is tempting, at this point, to hear the late Michael Winner saying: ‘Now calm down dear.’ 78
Thaler and Sunstein 2008, n 2 above, 37.
19
This accumulating of nudges, it could be argued, multiplies the undermining of
autonomy that even single (Second or Third Degree) nudges involve.
Here we see that a supposed argument for nudging may have force in relation to
one degree of nudge but not to others. Thus, when Thaler and Sunstein argue that
‘rare, difficult choices are good candidates for nudge’79
it may be conceded that some
positive potential is offered by a First Degree nudge that is designed to enhance
conscious decision-making by rendering complex data more manageable. It will be far
less readily acknowledged that there is case for a Second or Third Degree nudge
since, as just argued, this is likely to conduce to a cumulative undermining of
autonomy.
A further concern about Second and Third Degree nudges relates to their longer-
term and extensive use. It is that such responses do not seek to improve rational,
informed decision-making but serve to reduce the citizen’s voice in his or her destiny.
Used over time and numbers of issues, the effects of such accumulations of nudging
may be to produce control regimes that are defeatist about the capacities of
individuals to become more responsible and deliberative. The worry is that
proliferations of nudges produce citizens with ‘choice structuring fatigue’ and they
positively discourage the kinds of learning, responsible and thinking behaviour that
most governments espouse. Any short term positive gains from nudge may be
outweighed in time as people both ‘zone out’ from nudges and succumb to an
‘infantilisation effect’ by assuming less responsibility for their own welfare.80
It can be argued, furthermore, that many citizens may be unhappy about state
Second and Third Degree nudges in a way that they are not about commercial or other
decision-shaping pushes.81
There may be a number of reasons for this. They may see
commercial nudges, such as advertising campaigns and store designs, as consistent
and predictable facts of life that they are comfortable in dealing with. They accept that
salespersons will puff their products and engage in like behaviour but they may even
enjoy negotiating their way through the market, provided that no lies are told. In
contrast, they may see state nudges of the Second and Third Degree to be far more
covert, unknowable and threatening than the expected strategies of marketing and
sales. The public, moreover, may fear the scale of organized governmental nudging
much more than they would a random collection of market-driven nudges.
Expectations may also differentiate between nudges from private and public
sources. Citizens and consumers may anticipate and accept that the private sector will
behave in a profit-enhancing manner and use methods to further such ends that are not
wholly transparent. The public, though, may expect better things from the
governments that they have elected and entrusted with vast sets of powers. There is
likely to be a demand that, in using these powers to control and shape peoples’ lives
79
Thaler and Sunstein 2008, n 2 above, 82. 80
See Bovens n 14 above, 11. For similar concerns that ‘design’ or ‘code’ approaches to
regulation fail to respect citizens as morally responsible agents see Brownsword n 52 above
17-20 and D. Smith, ‘Changing Situations and Changing People’ in A. Von Hirsch, D.
Garland, A. Wakefield (eds.), Ethical and Social Perspectives on Situational Crime
Prevention (Oxford: Hart Publishing, 2000) 147-174 ( who notes the ‘demoralising effect’ of
regimes that remove personal choice). 81
Thaler and Sunstein note this worry (n 2 above, 11) but downplay it on the grounds that
nudges are more benign than commands and because nudges are inevitable. In portraying this
concern as a ‘non-starter’ Thaler and Sunstein do not take the point that nudges can be far less
transparent than commands. The ‘inevitability of the nudge’ contention also sidesteps the
point that nudges from some sources may be more acceptable than from others.
20
and decisions, ministers, officials and regulators use methods that are properly open
and accountable rather than covert and undermining of citizen autonomy. Nudges of
the second and third degree are liable to be seen as exploitative of citizens’
weaknesses and as ‘a form of disrespectful social control’.82
The public, on some
issues, will think: ‘If the government sees the need to control this issue, they should
tell us this and have the courage to seek parliamentary approval for new legislation
rather than operate with some sly nudge.’
A third defence of nudging would contend that the allegedly negative features of
nudge have to weighed against the behavioural and welfare gains that can be
achieved. This is a defence that is canvassed by some commentators but it is not one
that Thaler and Sunstein rely on heavily, since they take it that the heat is taken out of
this issue by their stipulation that opt-outs should be easy. 83
To balance the costs and benefits of nudging is, however, just one of a number of
ways of dealing with the case for nudging and two further approaches can be noted. A
second considers the proportionality of the nudge and a third looks to the public
acceptability of the nudge.
The essence of the cost-benefit approach is a comparison of the gains in health or
other aspects of welfare to be anticipated from a nudge and the losses to autonomy
and quality of life that a nudge may produce in the short or longer term.84
A
‘proportionality’ approach differs from a cost-benefit view in emphasising non-
quantitative and ethical factors. It adverts to such matters as the scale of the problem
to be addressed, the evidence that the nudge will be effective, and the representative
and moral considerations that are relevant. These latter will include such matters as
intrusiveness, restriction of freedom, quality of choices made, and transparency.85
A public acceptability approach asks whether the use of a nudge, in a given
context, is sanctioned by popular approval or ‘public permission’.86
For its part, the
House of Lords Science and Technology Select Committee was not convinced by this
latter approach and cautioned that: the public tended to resist change and its approval
might only be evident sometime after a nudge had been used. Popular approval tended
to vacillate over time, and the individuals making up the public might possess
distorted or misguided views on these matters for a number of reasons familiar to
behavioural economists.87
In the following section the approach that is adopted in assessing nudge comes
closest to the proportionality version. It draws out the messages from the above
discussion in seeking to identify the circumstances under which governments can be
expected to use nudge in a manner that is positive.
82
Hausman and Welch, n 14 above, 134. 83
Ibid. See, however, Thaler and Sunstein 2008, n 2 above, 249 where the authors admit that
they have no ‘inflexible rule’ on when opt-out costs are excessive and argue that: ‘The real
question is when we should be willing to impose some nontrivial costs in the interests of
improving people’s welfare.’ 84
The message of the behavioural approach that preferences are shaped, of course, renders
calculations of costs and benefits fraught because this calls into question the primary basis for
defining and assessing welfare – namely the satisfaction of individual preferences. (I am
grateful to Andrew Lang for making this point.) 85
See House of Lords Select Committee, 2011 para 2.19. 86
See House of Lords Select Committee, 2011 para 2.16. 87
ibid.
21
THE CONDITIONS FOR EFFECTIVE AND ACCEPTABLE USE
It is clear that First, Second and Third Degree nudges raise very different issues and
that this has to be borne in mind in looking at any potential nudge. Not only that but
different nudging tools have divergent potential in different contexts. This section
will, accordingly seek to deal with the different degrees of nudge and the different
tools for nudging whenever these present variations of challenge.
Can the conditions for effective use of nudge be stated? A first message from the
above discussion is that nudge cannot be expected to produce results effectively, as a
sole strategy for intervention, where the causes of the identified mischief go beyond
individual choice and relate to such matters as economic and cultural environments.
Secondly, it is a condition for effective use of nudge that the targets of the given
nudge are liable to be affected by the particular nudge tool employed. Table 2
indicated how dramatically reactions to nudges can be expected to vary across not
only targets with different characteristics, but also tool types, degrees of nudge, and
modes of implementing nudge tools.
Third, unanticipated side effects and unintended consequences have to be
anticipated as much with nudges as command rules, taxation mechanisms or other
intervention strategies. Similarly, it is necessary to anticipate the effectiveness
implications of any likely counter-nudges that may be forthcoming.
Fourth, mechanisms have to be adopted to deal with the danger that nudging
objectives lose contact with the relevant expectations of the public or the government.
This may demand that deliberative procedures are operated in parallel with nudges
and that feedback loops are established so as to inform and update the nudgers.
Fifth, the cumulative and long-term effects of nudging have to be taken on board in
order to avoid the tendency of short term gains to be achieved and the expense of
longer term, and greater, costs. Such effects include potential diminutions in trust,
nudge fatigue, and infantilisation.
Finally, a condition for the effective use of nudges across government is that
governments institute, on a broad basis, more studies that are designed to evaluate
behaviour change interventions.
Turning to the conditions for using a proposed nudge in an acceptable manner, the
central challenge is to satisfy expectations on representational and ethical fronts.
Regarding such matters, First Degree nudges can be separated out as far less
problematic than Second or Third Degree nudges. With a First Degree nudge, the
autonomy of the citizen is respected and conscious decision-making is enhanced. Few
representational or ethical problems can be expected to arise provided that the simple
information (or reminder) given is accurate. Some attention, nevertheless, may have
to be paid to the need to justify the decision to provide this piece of information or
this reminder rather than another.
Second Degree and Third Degree nudges are a different matter because they
detract from the autonomy of individuals to different extents. Acceptable use of nudge
demands that the nudgers have satisfactory responses to the following concerns. The
lack of public debate and accountability involved in many nudges. The imposition of
the nudger’s conception of welfare rather than the target’s. The limitations of opt-
outs, their variations in availability and practicality and their sometimes
discriminatory effects. A further concern that may call for a response is the fairness of
imposing on the autonomy and convenience of more sensible citizens in order to
assist those who are less responsible. With Third Degree nudges, the
emotional/psychological blocking of the opt-out is a special worry as is the removal
of opt-out through the re-shaping of preferences.
22
Justification may also be needed for choosing the nudge as a response to a
problem (even a decision-framing problem) rather than a command, economic
incentive or other intervention. The notion that nudges are inevitable, and that this
justifies governmental nudging, seems flawed on a number of counts.
How then, might one go about judging whether a proposed nudge would have
value in a given context? A starting point would be clarity on the character of the
mooted nudge. Is it a First, Second, or Third Degree example? Such characterisation
immediately throws challenges into relief. Next, it is essential to demonstrate with an
evidentially secure case that the outcomes of the proposed nudge are likely to be
secured effectively and that nudging will realise these in a superior fashion to more
traditional regulatory interventions.
On representational and ethical matters, one potential response to worries is to
show that processes ancillary to the nudge will ameliorate problems. Thus, references
can be made to transparency exercises such as disclosures of nudges or consultation
and feedback procedures. It can be expected, however, that, as noted above, general
strategies of disclosure will be seen as doing little to change the semi-covert nature of
many Second Degree nudges, and that disclosures relating to Third Degree nudges are
as likely to provoke protest as to reassure potentially targeted citizens.
A second response is to demonstrate that there are no representational and ethical
losses resulting from nudging since, in the given context, there is evidence that
individual autonomy is already so badly compromised that a nudge produces a
welfare enhancing result without further detracting from decision-making autonomy.
A difficulty here is to show that the alleged imperfections of autonomy that are cited
are not, in fact, merely manifestations of preference that the nudger lacks sympathy
with. A further problem with the ‘compromised autonomy’ argument lies in
contending that Second and Third Degree nudges will enhance decision-making rather
than simply place a further hurdle in the way of conscious decision-making.
Overall, the case has to be made that the gains from nudging will outweigh the
representational and ethical costs and worries noted above. The difficulty is that
Second and Third Degree nudges will routinely involve representational and ethical
costs that are both difficult to quantify and liable to be contested. A further
complication is that judgements will often have to be made about the acceptable
balance between short term gains and longer term losses.
In short, it is easier to identify the challenges and concerns involved in using
nudges than to quantify the prospects of making a good case for nudging in any single
context or more generally as a tool of governmental influence.
ANOTHER TOOL IN THE BOX?
At first glance, the safest conclusion to draw about nudge would seem to be that this
may not be the complete answer to governmental control but it is, at least, a very
useful device to have in the regulatory / non-regulatory toolbox of influences over
conduct. Further consideration, however, suggests that this conclusion builds too
readily on a particular conception of the toolbox of state control devices: one in which
different regulatory and non-regulatory tools take their places in an harmonious group
(a sort of seven dwarves choir of governance).
The reality is often that different such tools may compete rather than co-ordinate
harmoniously. In the first instance, the tools may vie for political prominence. Thus
the House of Lords Science and Technology Select Committee expressed the concern
in 2011 that the UK government’s promotion of behaviour change strategies was
23
blocking out consideration of other intervention strategies.88
This typifies the worry
that when politicians and ministers fall in love with a certain newly fashionable policy
style or mode of intervention, they tend to be blind to the virtues of the more
traditional approaches. This is a danger that can be responded to (at least partially)
with a First Degree nudge – as exemplified by the Science and Technology
Committee issuing a reminder to the Government that it should not forget that a range
of policy tools is available.
There is, moreover, a second more structural worry about seeing control tools as
an essentially harmonious package of options. Different regulatory and non-regulatory
tools of intervention operate with different logics in so far as they establish different
relationships with targeted persons and use varying ways of making demands of such
targets.89
The covert aspect of Second and Third Degree nudges involves little
dialogue between the nudger and the target. Indeed, as noted above, an openness of
dialogue can be expected to remove much of the force from certain types of nudge.
This contrasts with the position in more traditional regulation when steps are taken
routinely to clarify requirements, explain what compliance demands and so on. The
problem in combining intervention tools such as commands and nudges is that they
may undermine each other – it is difficult to establish a control relationship that is
based on openness, and, at the same time, to use nudges by stealth. The effect of
introducing nudging into an area may, thus, be to reduce the effectiveness and the
transparency of other modes of intervention and to detract from the broader regime of
control. This point was made above in noting that a problem of using nudges in the
medical field is that this can destroy the trust relationships that underpin the other
interactions between patients and physicians.90
A third problem in seeing the nudge as just another tool in the box is that choices
of different intervention tools are governed by different philosophies of control and of
state/individual relationships. It has been seen above that nudging operates on the
basis of libertarian paternalism so that flawed individuals’ decisions stand to be
corrected by the state so as to increase citizens’ welfare. With Second and Third
Degree nudges, the basis of the nudge is an exploitation of the limited cognitive,
volitional and emotional capacities of individuals. The idea in such paternalism, is not
to take steps in order to increase the quality of individuals’ rational, conscious
decision-making, it is to accept peoples’ flaws and use these to advantage.
The difficulty regarding the toolbox is that some tools of intervention operate on
quite different philosophical and ethical assumptions from those that underpin
nudge.91
Classical command and control rules of regulation, and economic incentives,
tend to build on the rationality of persons and firms – they issue prescriptions that are
overt because they take it that people will use their rationality to react positively to
the threats of sanctions or taxes. Enforcement practices tend to work on a rational
deterrence basis, and models such as the ‘responsive regulation’ approach, treat
regulated persons and firms as cognitively competent responders to overtly issued
messages.
The implication of these differences is that nudge does not sit wholly happily
alongside many other intervention tools. Choices have to be made about the position
88
House of Lords Select Committee, 2011 para 8.15. 89
See R. Baldwin and J. Black, ‘Really Responsive Regulation’ (2008) 71 Modern Law
Review 59-94. 90
Blumenthal-Barby and H. Burroughs, n 22 above, 5. 91
See the parallels with Brownsword, above n 52 on the differences between ‘East coast’ and
‘West coast’ intervention styles.
24
that the state wants to adopt regarding targeted individuals – is their autonomy to be
respected and their rationality to be openly worked with, or is their autonomy to be
reduced and their decision-making to be nudged by stealth? A decision by a
government to adopt one of these philosophies rather than the other involves a
disposition for or against using nudges as opposed to more traditional methods. For
this reason, a vision of the nudge as just another potentially useful intervention tool
may be one that is too easily arrived at.
This is not to argue that nudges have no role within government, it is to caution
that, when choosing when to nudge rather than intervene in another way, it is essential
to be clear and open about the philosophical basis for such a choice as well as to be
aware that different modes of intervention may operate with clashes of logic that
threaten not only effectiveness but also the serving of representative and ethical ends.
CONCLUSIONS
The nudge has burst on the scene in the last decade and it has proved enormously
alluring to certain governments, including those of the UK and USA. There are,
however, dangers in embracing the nudge uncritically. A first problem has been that
the nudge has often been discussed, at least in policy circles, with a lack of conceptual
precision. This article has sought to address this deficiency by distinguishing between
three distinct degrees of nudge and by exploring the different concerns that relate to
nudges of these three orders. This, at least, ensures that the case for nudging in a
given context will not be assessed on the assumption that all behaviour change
strategies are as benign as the First Degree nudges described here.
It has been seen that it is difficult to state, in precise terms, the conditions under
which nudging can be used effectively and acceptably. This is because there are
numbers of different nudging tools and three different degrees of nudging severity to
be taken into account alongside varieties of target. It has been possible, however, to
identify the challenges and concerns that are posed by nudges of different degrees and
to assess the extent to which these can be responded to in positive terms.
It should be emphasised, finally, that the above parade of reservations about nudges
does not prompt the conclusion that the nudge should be abandoned. The argument
here is that the nudge strategy must be used with precision and an awareness of both
its limitations and its fit within the range of state interventions. The three degrees of
nudge are here to stay and there will be more instances of behaviour change – one can
only ask of the nudge: ‘When will I see you again?’